Two essays on corporate default risk

I find strong evidence that higher state ownership leads to lower default risk due to soft budget constraints.. Using data from China, this paper provides strong evidence for the negativ

TWO ESSAYS ON CORPORATE DEFAULT RISK

2014

submitted fopreviously

_

4

d it has beenources of inf

s

ee in any un

n written byformation

niversity

y

Acknowledgements

I would like to express my deepest gratitude to Professor Anand Srinivasan,

my supervisor, for his patient guidance and enthusiastic encouragement throughout the Ph.D program Without his help and support, the completion

of this thesis would not have been possible My heartfelt thanks also go to the other two thesis committee members, Professor Joseph Cherian and Professor Charles Shi, who gave me valuable comments and insightful feedbacks for this thesis I am also very grateful to the research grant “Role of government in credit markets, No R-315-000-104-646” for providing the financial support for Chapter 1 Moreover, I would like to express my gratitude to the Risk Management Institute at NUS for providing me the comprehensive data used

in this thesis I also want to thank Professor Duan Jin-Chuan at RMI for providing many valuable comments

Besides, I am grateful to the professors in the Department of Finance and my Ph.D classmates, for their great help and support on my Ph.D journey during the past five years I want to express my apology for not listing each name of you here individually

Last but not least, I am deeply indebted to my parents Du Junlong and Shen Lingwan for their unconditional love and support They inspire me to be always positive, passionate and self-confident They let me know the meaning

of the life and the beauty of the life

Table of Contents

Acknowledgements III Summary VI List of Tables VII List of Figures VIII

Chapter 1: State Ownership and Firm Default Risk: Evidence from China 1

1.1 Introduction 1

1.2 Hypotheses Development 7

1.3 Some Background on Chinese SOEs 9

1.3.1 Overview 9

1.3.2 History of SOE reform 11

1.3.3 SOEs in other countries 13

1.4 Data and Summary Statistics 14

1.4.1 Data and sample selection 14

1.4.2 State ownership 15

1.4.3 Default events 17

1.4.4 Measures of default probability 18

1.4.5 Summary statistics and univariate analysis 20

1.5 Empirical Results 21

1.5.1 The predicting power of state ownership on corporatedefault events 22 1.5.2 The effect of state ownership when a firm is facing global negative industry shock 23

1.5.3 The reduction of state shares 25

1.5.4 The effect of state ownership under different industry competitiveness environments 26

1.5.5 The effect of state ownership when budget constraint becomes harder ……….27

1.5.6 State ownership and the probability of obtaining bank loans 28

1.5.7 Discussions on the dummy variable SOE 30

1.6 Conclusion 30  

Chapter 2: Firm Default Risk and Currency Return: An International Study 59

2.1 Introduction 59

2.2 Data and Main Variables 64

2.3 Summary Statistics 67

2.4 Empirical Models and Results 69

2.4.1 The prediction power of currency returns on firm default events 70

2.4.2 The effect of currency return and a country’s international trade 72

2.4.3 The asymmetric effects of currency return 73

2.4.4 The effect of currency return and a country’s exchange rate policy 75

2.4.5 The effect of currency return and financial market development 76

2.5 Conclusions 77

  Bibliography Bibliography 1 32

Bibliography 2 79

  Appendix Appendix 1: Variable Definitions 35

Appendix 2: State ownership and default ratio across different industries The industry classification is based on CSMAR Industry Name B 52

Appendix 3: Variable Definitions 81  

Summary

This thesis includes two essays on corporate default risk

The first essay directly tests the association between state ownership and firm default risk, using a sample of Chinese listed firms from 1990 to 2011 I find strong evidence that higher state ownership leads to lower default risk due to soft budget constraints State ownership has a stronger effect when firms are facing global negative industry return Moreover, the effect of state ownership will be more significant for firms operating in competitive industries Also, I find that state ownership has a less significant effect for firms located in regions with less government intervention and a better legal environment, where the budget constraint is harder

In the second essay, I find strong evidence for the prediction power of currency return on firm default risk And large local currency deprecation is a major reason for the positive association between currency return and default risk Using country-level international trade data (the sum of exports and imports) as proxy for the likelihood of using foreign currency debt, I find that currency return has a greater effect for countries that more rely on international trade, providing supporting evidence for the channel of foreign currency debt that connects the exchange rate and firm default risk Moreover,

I find that while large currency depreciation could lead to higher default risk, small depreciation is good for countries with trade surplus (exports are larger than imports) and small appreciation is good for countries with trade deficit In addition, the effect of currency return is less significant for countries with restrictions on exchange rate and less significant for countries with better financial market development

List of Tables

Chapter 1

Table 1.1 Summary Statistics 38

Table 1.2 Spearman rank correlation (p- value in parentheses) 45

Table 1.3 The default probability of SOEs 46

Table 1.4 The effect of state ownership when firms are facing negative global industry shock 47

  Table 1.5 The firm characteristics before the reduction of state shares 48

  Table 1.6 The effect of state ownership under different industry competitiveness environments 49

  Table 1.7 The effect of state ownership under different market development environments 50

  Table 1.8 The effect of state ownership on the probability of getting bank loans 51

Chapter 2 Table 2.1 Summary Statistics 85

Table 2.2 Default Ratio Distribution 86

Table 2.3 Default ratio summary for each economy (United States excluded) 87

Table 2.4 The prediction power of currency return on firm default event 89

Table 2.5 The prediction power of large currency depreciation on corporate default event 90

  Table 2.6 The effect of currency return and the international trade 91

Table 2.7 The different effects of currency return when countries are in trade surplus or deficit 92

  Table 2.8 The effect of currency return and country exchange rate policy 93

Table 2.9 The effect of currency return and financial market development 94  

Chapter 1: State Ownership and Firm Default Risk:

Evidence from China

1.1 Introduction

Reporting on the Yunwei Co., Ltd., a manufacturing company in China, the

Financial Times, Asia Edition, August 28, 2013, noted that:

“It (Yunwei) lost Rmb 1.2bn ($196m) last year, at times using just two-thirds

of its production capacity….As things deteriorate, Yunwei at least has a cushion to fall back on Its parent company is owned by the Yunnan provincial government, and officials in China have shown repeatedly that they are extremely reluctant to see their local champions fail….”

Financial Times, Asia Edition, August 28, 2013

The author of this article clearly expresses his view that the government will provide guarantees to state-owned enterprises (SOEs), a view widely accepted

by the public and assumed in many studies However, the relationship between firm default probability and state ownership has not been directly examined in academia, although we can see some hints or indirect evidence from past studies Using data from China, this paper provides strong evidence for the negative association between state ownership and firm default risk, and endeavors to help us better understand the roles of government, competitions and market development in the economy

Throughout history, politicians and economists have debated the role of government in the economy The mass of previous literature examined the effectiveness of state ownership and private ownership, providing strong empirical evidence for the advantages of private ownership (see Eckel and Vermaelen, 1986; Chen, et al., 2008; Firth, et al., 2010; etc.) Moreover, many studies show that there is significant improvement in operating performance or equity value after privatization (Megginson and Netter (2001) summarize earlier findings; Sun and Tong, 2003; Megginson, et al., 2004; Boubakri, et al., 2011; etc.) However, the impact of state ownership on default risk has not been investigated

The objective function that the government faces differs from that of private investors The government might need to maximize social welfare, maintain a high employment rate, improve education and infrastructure, maintain the stability of society, and provide support to some industries of strategic importance to the country SOEs play a crucial role for the government to achieve these goals Thus, the government is reluctant to allow these firms to default and might provide guarantees for SOEs This phenomenon is known as

a soft budget constraint, a term first introduced by Kornai (1979, 1980, and 1986) Kornai and many other economists believe that the soft budget constraint arises from various state-imposed policy burdens and is the major source of inefficiency for firms in socialist economies (Lin, et al., 1998; Berglof and Roland, 1998; and Frydman, et al., 2000; etc.) In addition, some studies suggest that capitalist economies also have the soft budget constraints (Maskin, 1999; Kornai, et al., 2003) Government subsidies, soft taxation, soft credit and soft administrative prices are all means to soften the budget

constraint Cull and Xu (2003) examine the two major methods of government bailout in China from 1980-1995: direct government transfers and loans from state-owned banks They suggest that the bailout responsibilities were increasingly imposed on banks after 1990 Moreover, some studies provide indirect evidence for the soft budget constraint by comparing some characteristics of SOEs and non-SOEs For example, Acharya and Kulkarni (2012) show supporting evidence that state-owned banks have access to stronger government guarantees and forbearance, by examining the deposit and lending growth of banks during the financial crisis Borisova and Megginson (2011) and Borisova, et al (2012), find that state ownership leads

to lower cost of debt during the financial crisis due to the guarantee effect, using the European privatization and government investment sample, respectively Therefore, due to the existence of a soft budget constraint, companies with higher state ownership might have a lower default risk

However, conversely, a soft budget constraint might worsen the moral hazard, increase the agency cost, lead to lower firm value, and thus lead to a higher risk of default Managers might not focus on firm value maximization, and instead will try to find the cash and credit subsidies from the government, and might give priority to the social and political goals of the government Furthermore, state ownership provides a lower level of monitoring and the government guarantees also remove the monitoring incentive of other stakeholders (Bortolotti, et al., 2010) Also, the presence of a soft budget constraint will affect the firm’s investment behavior SOEs might take more risky investment and have lower investment-cash flow sensitivities (Chow, et

al., 2010) Therefore, the agency costs arising from the soft budget constraint might lead to a higher default risk

Therefore, empirical investigation is needed for the association between state ownership and default risk due to the direct soft budget constraint effect and the agency cost effect arising from the soft budget constraint In this paper, I present empirical evidence that state ownership leads to lower default risk, using Chinese listed firms’ data from 1990-2011 I find strong predicting power of state ownership on firm default events after controlling several popular measures of default risk These measures of default risk include Altman’s (1968) Z-Score, Merton’s (1974) Distance-to-Default (DTD), and the Probability of Default (PD) of Duan, Sun and Wang (2012), which mainly incorporate firm’s financial and market information I also test the effect of state ownership when a firm is facing global negative industry return, which can be viewed as an exogenous shock to the firm I find that state ownership has a more significant effect on default risk during the shock period This shock can be used to address potential endogeneity problem

To examine whether the negative association between state ownership and default risk is only driven by some SOEs in natural monopoly industries, I conduct regressions using different subsamples based on industry competitiveness I find that the effect for state ownership is more significant for firms in competitive industries This finding helps differentiate the effect

of state ownership with the effects of natural monopolies

I also test the effect of state ownership when the budget constraint becomes harder I find that state ownership has less effect for firms located in areas

with a better legal environment and less government intervention The index

of legal environment and government intervention is from Fan and Wang (2011) The results suggest that the effect of state ownership on default risk is less significant when the budget constraint becomes harder Moreover, using the data of bank loans from China Stock Market & Accounting Research (CSMAR), I test one channel of the soft budget constraint I find that firms with state ownership are more likely to get loans from state-owned banks

There are several reasons why I use Chinese data in this study First of all, state ownership is still very popular among Chinese firms and more than 60%

of listed firms in China are SOEs According to an Organisation for Economic Co-operation and Development (OECD) study by Christiansen (2011), there are only 48 listed SOEs among 27 countries Thus, China-listed firms provide

a large sample for analysis Secondly, Chinese SOEs cover almost every industry sector, whereas among the 27 OECD countries, almost 75% of SOEs are in the utilities and financial sectors Firms in the utility sector are probably natural monopolies and financial institutions play a special role in the economy Thus, using Chinese data, it is possible to examine the state ownership effect and to avoid the natural monopoly effect and the financial sector effect Thirdly, in the geographic dimension, there are significant differences among different regions in terms of market development Thus we can examine the effect of state ownership under different legal environments and market development levels This helps us better understand the role of government and the role of the market

This paper contributes to the literature on government guarantees In previous studies, it is assumed that government provides a guarantee to SOEs and is

reluctant to allow SOEs to default Although this view is widely accepted by the public, the direct empirical evidence is missing This paper is the first to directly test the effect of state ownership on default risk, and provide evidence that the presence of state ownership leads to lower default probability The finding could help us better understand the role of government in the economy Moreover, this paper makes contributions to the default forecast literature Previous default forecast models mainly incorporate a firm’s financial and market information This study suggests that the ownership structure, which might affect firm value over a longer period of time, should also be incorporated into the forecast model, at least into the forecast model with the longer time window

Most previous studies on state ownership focus on effectiveness, and only two papers (Borisova and Megginson, 2011; Borisova, et al., 2012) examine the association between tate ownership and cost of debt, areas which are the closest to this study This paper differs from their studies in several aspects Firstly, the samples are different Borisova and Megginson (2011) use the European privatization sample, and Borisova, et al (2012) use the European government investment sample Nearly 60% of the observations in Borisova and Megginson (2011) are for banks, and 34% of the investment deals are in the financial sector in Borisova, et al (2012) My sample includes all the Chinese listed firms with data available on CSMAR and the National University of Singapore Risk Management Institute (NUS-RMI) database (NUS-RMI, 2013) And only 29 firms are in the financial sector Because of the different economic roles of financial firms and non-financial firms in society, they should have different abilities to access government guarantees

Secondly, our findings are different Borisova and Megginson (2011) find that state ownership leads to lower cost of debt, but fully privatized firms (zero state ownership) have lower cost debt compared to partially privatized firms And Borisova, et al (2012) find a negative association between state ownership and cost of debt only during a financial crisis period The results in this paper suggest a linear relationship between state ownership and default risk

The remainder of the paper is organized as follows Section 2 develops testable hypotheses Section 3 introduces some background on Chinese SOEs Section 4 describes data and summary statistics Section 5 performs and discusses empirical analyses Section 6 concludes

1.2 Hypotheses Development

It is widely accepted by the public that the government will provide guarantees

to SOEs and is reluctant to allow SOEs to default This phenomenon is referred to as soft budget constraint, in many studies The motivation for the government is to achieve its social and political goals, such as maintaining the employment rate, improving education and medical services, supporting industry sectors of strategic importance to the safety of the country Government guarantees through bank loans, fiscal subsidies, and soft taxation might lead to lower default risk However, on the other hand, the presence of soft budget constraints might worsen the managerial moral hazard and increase agency costs The corporate governance problem arising from soft budget constraint might increase the firm’s default risk Thus, the relationship between state ownership and default risk is still an empirical question In

China, the legal system is not well developed, and a modern corporate governance scheme has yet to be established in both SOEs and non-SOEs Many non-SOEs are family-owned firms, and might suffer more severe moral hazard problems Thus, the government guarantee effect might be more significant than the agency cost effect arising from the soft budget constraint, for Chinese firms We could expect that firms with state ownership have lower default risk

Hypothesis 1 (H1): The presence of state ownership leads to lower

probability of default

I can conduct a test to examine the effect of state ownership when the firm is facing global negative industry return, which can be viewed as an exogenous shock to the firm If the negative association between state ownership and default risk is due to the soft budget constraint, we could expect that the effect

of state ownership will be stronger during the shock period This shock to default risk can be used to deal with potential endogeneity problem

Hypothesis 2 (H2): State ownership has more significant effect on

default risk when firms are facing global negative industry shock

Since many SOEs are in concentrated industries such as utilities, natural resources and telecommunications, the negative association between state ownership and default risk might be driven by those SOEs To differentiate the government soft budget constraint effect and the natural monopoly effect, we can test the relationship using different subsamples based on industry competitiveness And due to the soft budget constraint, we could expect that state ownership still would have a significant effect for firms in competitive

industries Furthermore, because of the strong competition, SOEs operating in competitive industries are more likely to acquire government guarantees, and the state ownership effect will be stronger in competitive industries

Hypothesis 3 (H3): The effect of state ownership is more significant for

firms in competitive industries

Moreover, in an environment with a better legal system and less government intervention, the budget constraint will be much harder Fan and Wang (2011) provide a marketization index for China’s provinces, which is widely used in research on China Among the 23 indicators of the comprehensive index, there

is one indicator for the legal environment, and another one for government intervention The two indicators are based on the survey of more than 4,000 enterprises in China Using the two indicators, we can test the effect of state ownership under different legal and market environments; and, we could expect that the effect of state ownership is less significant for firms located in regions with a better legal environment and less government interventions

Hypothesis 4 (H4): The effect of state ownership is less significant for

firms located in regions with a better legal environment and less government intervention

Commission (SASAC) of the State Council, provincial SASACs and municipal SASACs, China Banking Regulatory Commission (CBRC), China Insurance Regulatory Commission (CIRC), China Securities Regulatory Commission (CSRC), or government ministries such as Ministry of Commerce, Ministry of Education; 2) enterprises effectively controlled by SOEs or their subsidies; 3) urban collective enterprises and village enterprises Usually, the first two categories are considered as SOEs Central SOEs include entities managed by SASAC of the State Council; state-owned financial institutions supervised by CBRC, CIRC, CSRC; entities owned by central government ministries When China was a centrally planned economy, SOEs were fully owned by the state Nowadays, the SOEs refer to state-owned and state-holding enterprises After nearly 35 years of privatization, restructuring, joint ventures, mergers and acquisitions, the ownership structure of SOEs has become much more complicated, and thus it is difficult to clearly define the state shares and to provide accurate statistics on SOEs According to OECD (2009), a study of Chinese SOEs, it is difficult to find a consistent data set that could distinguish between state-owned and non-state-owned legal entity shares And based on this study, 70% of listed Chinese non-financial firms are SOEs in 2004, by identity of the largest shareholders By the end of 2008, there are 149 central SOEs controlled by SASAC of the State Council, and the subsidies of these central SOEs might exceed 10,000

The contribution of SOEs on gross domestic product (GDP) is large Based on

a report for the U.S.-China Economic and Security Review Commission performed by Szamosszegi and Kyle (2011), SOEs accounted for 45% of non-agricultural GDP and 40% of GDP in 2007 For employment, pure SOEs

(fully owned by the SASACs or government ministries) account for nearly 30% of the urban employment rate in 2009, based on the National Bureau of Statistics of China Although there is a clear diminishing trend of SOEs’ contribution, SOEs still remain a significant component in the economy

Strategic industries, which are important to China’s economic and national security, including defense, electric power and grid, petroleum and petrochemical, telecommunications, coal, civil aviation and shipping, are wholly or largely controlled by the state For some other important industries, so-called pillar industries, including equipment manufacturing, auto, information technology, construction, chemicals, iron and steel, non-ferrous metals, and surveying and design, the state holds significant ownership, not majority ownership For historical reasons, SOEs still exist in other industries, such as food and beverage, hostel SOEs are present in almost all the industries

The government maintains significant influence over SOEs The government decides on the appointments of top executives of SOEs and on their future career paths after leaving the SOE Thus, the executives of SOEs have strong incentives to follow the government’s policy and to achieve the social and political goals of the government SOEs, as an instrument of government policy, play significant roles in technology innovation (high speed rail), importing raw materials from other countries, and will continue their important role in the Chinese economy

SOE reform since 1978 can be divided into two stages:

Stage 1: 1980s and early 1990s Prior to 1978, the government determined the production level of SOEs SOE reform was focused on revitalization

by giving incentives and providing managers with more decision-making power At this stage, the SOEs had more flexibility in production and could make adjustments to their production plans based on market information Moreover, SOEs started to establish the Manager Responsibility System in the late 1980s Under this system, the manager took full responsibility for the SOE’s operation and the government should not intervene in the SOE’s decision making However, at this stage, there was no significant change in the ownership structure and governance structure, and the low efficiency problem had not been solved SOEs’ profitability was decreasing in the late 1980s and early 1990s According

to OECD (2009), in 1997, 6,599 companies out of about 22,000 large- and medium-sized SOEs recorded losses SOE reform became a priority for the premier, Zhu Rongji

Stage 2: Since 1997, when Zhu Rongji became the premier of China First

of all, the government realized that it could not manage so many SOEs, and therefore adopted the strategy “Zhua Da Fang Xiao” (Keep the larger SOEs, release the smaller SOEs) The smaller SOEs were allowed to go bankrupt, to be acquired or become privatized Secondly, to enhance SOE performance, the government implemented strategies such as huge layoffs, debt reduction, and technology improvement Thirdly, four Asset Management Corporations were established to deal with the bad loans of the four largest state-owned banks By the end of 2001, 4,000 out of 6,599 money-losing SOEs earned positive net profits (OECD, 2009) At the same

time, the ownership structure and corporate governance structure started to reform According to OECD (2009), the SOEs began to establish the

“Modern Enterprises System”: 1) clarification of property rights; 2) clarification of rights and responsibilities; 3) separation of politics and business; and 4) scientific management SOEs were encouraged to be listed in stock exchanges and raise capital from the public

According to an OECD study conducted by Christiansen (2011), there are only

48 listed SOEs in 27 OECD countries In terms of sectoral distribution, most

of the listed SOEs are in the utilities sectors, while some are financial institutions In fact, due to the financial crisis, Germany and the United Kingdom have become minority owners of large financial institutions Only Finland, France, Italy, Norway and Poland maintain minority state ownership

in listed manufacturing companies Around half of all SOEs, including listed SOEs, are in the network sectors (transportation, power generation and other energies) Financial institutions account for one-fourth of SOEs’ total valuation For some Scandinavian nations or countries that have recently made

non-a trnon-ansition townon-ards mnon-arket economies, such non-as the Czech Republic, Finlnon-and, Israel, Poland and Norway, SOEs account for 20% to 30% of the GDP On average, for the 27 countries studied in Christiansen (2011), SOEs account for 15% of the GDP

Also according to Christiansen (2011), for the 27 OECD countries, there are two types of state-owned shares: those directly held by the state; those held by

state-controlled financial institutions such as government-owned insurance and pension schemes, and government-owned investment funds

1.4 Data and Summary Statistics

The sample includes all listed firms in China’s Shanghai Stock Exchange and Shenzhen Stock Exchange with relevant data available in CSMAR, a widely used Chinese financial database, and the NUS-RMI database The Credit Research Initiative database of the Risk Management Institute (RMI) of the National University of Singapore provides several measures of default probability, such as Probability of Default (PD) and Distance-to-Default (DTD) Moreover, the RMI database also provides comprehensive information

on both market data and financial data on about 60,000 exchange-listed firms

of 106 economies around the world I retrieved the data used in this paper from the RMI database in January 2012 For those firms with both A shares and B shares traded on the stock exchange, I only include observations for A shares I obtain the state share data and the firm ultimate controlling shareholder data from CSMAR After the split share reform introduced in

2005, non-tradable shares become tradable, but the tradable state shares are not recorded in CSMAR Thus, only the state share data before the split share reform are used for analysis I also define SOE based on the type of firm ultimate-controlling shareholder I merge the ownership data with the RMI database, and obtain around 14,000 firm-year observations for regression analysis

1.4.2 State ownership

China’s two stock exchanges, the Shanghai Stock Exchange and the Shenzhen Stock Exchange, were established in 1990 There are two types of shares traded on these exchanges: A shares (RMB-denominated) and B shares (foreign currency-denominated) Under the split share structure, established from the beginning, A shares are further divided into tradable shares and non-tradable shares In 1990, approximately two-thirds of the A shares are non-tradable shares The two major holders of non-tradable shares are the state (government departments and agencies) and legal entities (the underlying companies and executives) (Guo and Keown, 2009) In April 2005, the Chinese government initiated a split share structure reform to convert all non-tradable shares into tradable shares By the end of 2007, the reform was complete for most companies, which represent over 97% of the total A-share market capitalization (Li, et al., 2011)

In this paper, I construct two variables for state ownership The first is a

dummy variable, SOE A firm is defined as SOE if the ultimate controlling

shareholder is: 1) SASAC of the State Council, provincial SASACs or municipal SASACs; 2) CBRC, CIRC, or CSRC; 3) government ministries; 4) Other SOEs The ultimate controlling shareholder information is available on the CSMAR database and is extracted from firm annual reports The yearly data are available from 2003 to now Since the listed firms usually are large and there is almost no complete privatization of large SOEs before 2003, I assume for years before 2003, the ultimate controlling shareholder is the same

as that in 2003 I also define the central SOEs based on the ultimate controlling shareholders CSMAR’s definition of controlling shareholder is

based on CSRC’s Administration of Takeover of Listed Companies Procedures That is, a shareholder is classified as controlling shareholder if he satisfies any one of the following scenarios: 1) the one holds more than 50%

of the total shares; 2) the one who holds more than 30% of the voting rights; 3) the one who can decide the appointments of over half of the board directors in

a listed company The ultimate controlling shareholder is the last layer of the shareholding relation chain

The other state ownership variable is the percentage of state shares, defined as the ratio of the number of state-owned shares divided by the total number of shares The state-owned shares are non-tradable shares owned by the state After the split share structure reform, the non-tradable shares become tradable shares, and many shares are owned by other state-owned companies, making the ownership structure much more complicated Thus the state share is very difficult to define clearly In my sample, only the state share data before the reform are included

Panel A of Table 1.1 reports the summary statistics for SOE The percentage

of SOEs is more than 60% for the years before 2009 Then it decreases to around 40% after 2009, probably due to the state share reduction in the split share reform and the state share transmission reform starting from 2009 The statistics are similar to OECD (2009) Panel B reports the summary statistics for state shares Approximately, 70% of the total companies have state-owned shares The mean of state shares is in the range of (0.265, 0.360) The mean of the state shares of the whole sample is 35.1%, from Panel E, and the standard deviation is 26.1%, statistics almost the same as those in Li, et al (2011)

Among all the observations, 25% are below 3.6% and almost 50% are above 40%

The first variable SOE, the dummy variable, is defined based on the control

rights, while the second one State Shares uses ownership data To investigate the correlation of the two variables, I examine the state share distribution of SOE sample and non-SOE sample The results are presented in Panel C and

Panel D The mean of state shares of SOE sample (SOE = 1) is 43.2%, while

the mean of non-SOE sample is 18.4% Panel C also reports the mean of state shares by year In most years, the mean of state shares for SOE sample is higher than 40%, which is much higher than non-SOE sample From 1997, the difference between the two samples is increasing, probably due to the SOE reform “Zhua Da Fang Xiao” (Keep the larger SOEs and release the smaller SOEs) Panel D reports more details for the comparison For SOE sample, almost 75% of the observations have more than 30% state shares Among the observations of non-SOE sample, 50% are below 4.2%

The dependent variable in the main regressions is Default, a dummy variable indicating the happening of default events The default events are extracted from the RMI database These events are collected from many resources, including Bloomberg, Wind Financial database, Compustat, The Center for Research in Security Prices (CRSP), Moody’s reports, Taiwan Economic Journal (TEJ), exchange web sites and news sources A challenging problem is that the definition of default might vary across different data sources RMI applies a default definition consistently across different economies Based on

the RMI technical report (2013), the default events can be classified under one

of the following events:

1 Legal impasse to the timely settlement of interest or principal payments, such as bankruptcy filing, receivership, administration, liquidation;

2 Missed or delayed payments of interest or principal, not including delayed payments made within a grace period;

3 Debt restructuring or distressed exchange, in which a new security

or package of securities is offered to debt holders, resulting in a diminished financial obligation (such as a conversion of debt to equity, debt with lower coupon or par value, debt with lower seniority, debt with longer maturity)

The main control variables used in my analysis are several popular default risk measures from previous default risk models:

Z-Score: Altman’s Z-Score is calculated by the following equation:

jt k

for a developing country (China in this paper), the variable SL/TA, is not used Distance-to-Default (DTD): Based on Merton’s (1974) Distance-to-Default model, distance-to-default measures the distance between the current value of assets and the debt amount in terms of asset volatility It can be calculated as the following:

)(

)))(

2/(()

t T

t T L

V

DTD

V

V t

where Vt denotes the current value of assets, L denotes the liabilities, and

is the asset volatility These data are available in the RMI database Duan and Wang (2012) discuss the estimation methods for DTD calculation For financial and properties firms, which typically have higher leverage, the KMV Corporation’s estimation seems ill-suited Thus, a transformed-data maximum likelihood estimation (MLE) approach is applied to the DTD calculation in the RMI database

Probability of Default (PD): Duan, et al (2012) propose a forward intensity approach for the prediction of corporate defaults over different future periods And the prediction is very accurate for short periods, with the accuracy ratios exceeding 90% for 1- and 3-month horizons and 80% for 6- and 12-month horizons using U.S data The accuracy ratio decreases when the horizon is increased to two or three years, but its performance remains reasonable This measure incorporates the profit, liquidity and market information of the firm The data are available on the RMI database The PD for a 1-year horizon is used in this paper

V



1.4.5 Summary statistics and univariate analysis

Table 1.1 Panel E reports the summary statistics for the variables in the main

regressions All the variables except SOE, State Shares, Default and PD are

winsorized at 1% The default ratio is calculated as the number of defaults divided by the total number of firm-year observations The default ratio for the whole sample is 2.0% Based on Panel F, the default ratio for SOEs is 1.4%, while non-SOEs’ default ratio is 2.6% Non-SOEs have a significantly higher default ratio

I divide the whole sample into quartiles Q1 to Q4 based on the state shares Q1 represents the quartile with smallest state share, and Q4 represents the largest state share quartile Panel G of Table 1.1 reports the default ratio of these four subsamples The default ratio of Q1 is 0.036, which is significantly higher than that of Q4 (0.01), suggesting that firms with lower state ownership have a larger likelihood to default The t-statistic for equality test (Q1 vs Q4) is 5.95, which is significant at the 1% level

Panel H describes the state ownership and default ratio in terms of industry sectors The sample covers almost all the industry sectors, and only 29 financial firms (1.4% of the total number of firms) are included The Properties sector has the second lowest percentage of SOEs (52.7%), and the highest default ratio (4.1%, much higher than the average 2%) This table also suggests a negative association between state ownership and default risk Appendix 2 reports the state ownership and default ratio in terms of a much narrower industry classification

[Insert Table 1.1 Here]

Table 1.2 presents the Spearman rank correlation matrix The correlation

between Default and State Shares is -0.061, and is significant at the 1% level,

suggesting a negative association between state ownership and default risk

The correlations between Default and PD, Z Score and DTD are 0.158, -0.160

and -0.071, respectively, and all are significant at the 1% level, suggesting that these several previous measures of default probability work very well These measures capture the firm’s liquidity, profit, competitive position in the industry and market information, and will be the main control variables in the following regression analysis

[Insert Table 1.2 Here]

1.5 Empirical Results

In this section, I first test the predicting power of state ownership on corporate default events To address the potential endogeneity problem, I examine the effect of state ownership when firms are facing shocks on default risk Moreover, the effect of state ownership under different industry competitiveness is investigated I also examine the effect of state ownership for the firms located in regions with a better legal environment and less government intervention, when the budget constraints become harder In addition, I use the data on bank loans from CSMAR to test whether firms with state ownership could more easily obtain loans from banks or state-owned banks This test could provide evidence for one of the channels of soft budget constraints

1.5.1 The predicting power of state ownership on corporate default events

Using probit regressions, I test the predicting power of state ownership on a corporate default event I employ the following yearly regression model:

Default it+1 = δ0 + δ1StateOwnership it + δ2ZScore it + δ3DTD it + δ4PD it

+ δ5 Other Controls + Fixed Effect +e1 it ,

where the dependent variable Default it+1 is a dummy variable indicating the presence of corporate default events in year t+1 The coefficient on state ownership is expected to be negative due to the soft budge constraint effect, which suggests the government will provide a guarantee for firms with state ownership

The regression results are reported in Table 1.3 I include industry fixed effects and year fixed effects for all four regressions In column (1), after

controlling for PD, the coefficient on SOE is -0.240 (z-statistics = -3.93),

which is negative and significant at the 1% level, providing evidence for Hypothesis 1 that the presence of state ownership leads to lower default risk

In column (2), after controlling PD, DTD and Z Score, the coefficient on SOE

is still negative and significant (-0.254, z = -3.55) I include SOE_Central in

column (3), but the coefficient is not statistically significant This suggests that central state ownership does not have a stronger effect on firm default risk The possible reason might be that local SOEs still could access local government guarantees, and thus there is no significant difference in default risk between central SOEs and local SOEs In column (4), I add more control

variables, several firms’ standard financial variables, such as Size,

Market-to-book ratio, Profit margin, ROA, Leverage The coefficient on SOE (-0.354 z

= -4.55) is still negative and significant There is almost no change in the

coefficients on SOE when I use different control variable sets

As expected, the coefficient on PD is positive and significant for all regressions, and the coefficient on DTD is negative and significant Since the

PD, Z Score and DTD have included the information on firm’s liquidity, profit

and market returns, the coefficients on many financial variables in column (4)

are not significant In the following regressions, I only include PD as the main

control variable

[Insert Table 1.3 Here]

1.5.2 The effect of state ownership when a firm is facing global negative industry shock

The global negative industry return can be viewed as an exogenous shock to firm default risk If state ownership does have an effect on the default risk, we could expect that the effect will be stronger during the negative industry shock period This also can address potential endogeneity problem caused by some unobservable variables In particular, state ownership and default risk might be both determined by some unobserved firm or industry characteristics For example, natural monopolies usually have lower default risk by nature, but at the same time they also have higher state ownership

I define the negative industry shock as an event when the industry return for the last year is smaller than -10% The industry return is calculated as the mean of stock returns of all the firms from 30 economies of the world, with

the data available on the NUS-RMI database The database covers the major economies from North America, Europe and Asia, such as the U.S., U.K., China, Japan, Germany, and France The industry is defined based on the Bloomberg Industry Subgroup Classification Since the industry return is calculated globally, the event can be viewed as an exogenous shock to firm default risk to the specific firm

Table 1.4 reports the regression results The sample of column (1) only includes observations with a negative industry shock, and the sample of column (2) includes observations without negative shocks I use the dummy

variable State_Dummy to indicate the presence of state ownership (equals 1 when the state share is not 0) The coefficient on State_Dummy is negative and

significant at the 1% level (-0.503, z = -2.72) However, the coefficient on state ownership is not significant in column (2) The combination of columns (1) and (2) shows the supporting evidence for Hypothesis 2, that the effect of state ownership is stronger when firms are facing shocks on default risk Using

the whole sample, I add the interaction term State_Dummy*Negative Industry

Shock in column (3) The Negative Industry Shock is a dummy variable

indicating the presence of the industry shock The coefficient on this interaction term is negative and significant (-0.402, z = -1.78), supporting Hypothesis 2

[Insert Table 1.4 Here]

1.5.3 The reduction of state shares

There is possibility that the government can choose the companies which have lower default risk To address this concern, I investigate the motivation of the government behind the events of the reductions in state shares

First, I construct a reduction sample including the events that there is a reduction in state shares of a company in a year Then I compare some firm-specific characteristics before the reduction in state shares of the reduction sample and the whole sample The results are reported in Table 1.5 It shows that the government reduces the state shares of smaller SOEs, even when the

smaller SOEs have larger profit margin and higher ROA The mean of Size of the reduction sample (Size=20.92) is significantly smaller than that of the whole sample (Size=21.31) The reduction might be because the SOE reform

“Zhua Da Fang Xiao” (Keep the larger SOEs and release the smaller SOEs) started by Premier Zhu Rongji

For the reduction sample, I further divide it into three subsamples: 1), PD increases by more than 10% after the reduction; 2), PD decreases by more than 10%; 3), the change in PD is less than 10% Then I compare the firm

characteristics before the reduction of the subsample 1) and 3) Table 1.5 also

reports this comparison The difference in Size is significant, suggesting that

larger firms are more likely to have a decrease in default risk Before the reduction, the firms with higher past stock return, higher Market-book ratio

and lower leverage are more likely to have an increase in PD This might be

because the firms with higher past stock return possibly will have lower return

in the future In general, the findings do not support the argument that the government will choose the companies with lower default risk

[Insert Table 1.5 Here]

competitiveness environments

Many SOEs are in natural monopoly industries or the financial industry Due

to the special roles of natural monopolies and financial institutions in the economy, they have more access to government guarantees Thus, the negative association between state ownership and default risk might be driven by the natural monopoly effect or the financial sector effect, not the soft budget constraint effect To address this concern, I examine the effect of state ownership under different industry competitiveness environments

There are two interesting questions here First of all, does state ownership still have an effect on default risk in competitive industries? If the negative association is driven by the SOEs in concentrated industries, state ownership will not have an effect in competitive industries Secondly, does state ownership have a more significant effect on default risk in competitive industries? SOEs operating in competitive industries face stronger competition from private firms, and they are more likely to acquire government guarantees Thus, we could expect that the effect of state ownership should be stronger for firms in competitive industries

I use two definitions for competitive industries: 1) HHI is smaller than the median value; 2) HHI is smaller than 0.15 The industries are defined based on

CSMAR Industry B classification (166 industry sectors) HHI is the

Herfindahl-Hirschman Index (HHI), defined as the sum of the squares of the

market shares of the firms in the same industry Table 1.6 presents the analysis results Column (1) uses the competitive industry sample based on definition

1 Column (2) uses the non-competitive industry sample based on definition 1 Based on definition 2, the competitive industry sample is used in column (3) and the non-competitive industry sample is used in column (4) The coefficient

on SOE for column (1) is -0.282 and significant at the 1% level But the

coefficient in column (2) is not significant, suggesting that the effect of state ownership is stronger for competitive industries and providing evidence for

Hypothesis 3 Similarly, the coefficient on SOE in column (4) is not

significant, while the coefficient is negative and significant in column (3)

[Insert Table 1.6 Here]

becomes harder

Budget constraints are much harder for firms located in regions with less government intervention and a better legal environment The indicators of government intervention and legal environment are from Fan and Wang (2011),1 and the two indicators are based on the survey of more than 4,000 enterprises in China Using the two indicators, I test the effects of state ownership under different legal and market environments and the results are

presented in Table 1.7 GovIntervention it is a dummy variable, which equals 1 when the government intervention index for the region is greater than the median value of all the regions at year t In column (1), the coefficient on the

interaction term SOE*GovIntervention is positive but not significant

However, in column (3), using the state share variable, the coefficient on the

interaction term State*GovIntervention is 0.512 and statistically significant,

suggesting that state ownership has less effect on firms located in regions with less government intervention Moreover, the coefficient on the interaction

term SOE*LegalEnviron is 0.256 and significant in column (2) In column (4), the coefficient on State*LegalEnviron is 0.650 (z = 2.17), which is positive and significant at the 5% level LegalEnviron is a dummy variable, which

equals 1 when the legal environment index for the region is greater than the median value of all the regions at year t This table reports evidence for Hypothesis 4

[Insert Table 1.7 Here]

BankLoan it+1 = δ0 + δ1StateOwnership it + δ2 Other Controls + Fixed Effects + e2 it ,

StateBankLoan it+1 = δ0 + δ1StateOwnership it + δ2 Other Controls + Fixed Effects + e3 it,

where BankLoan is an indicator of obtaining loans from banks and

StateBankLoan is the indicator of obtaining loans from state-owned banks

Since state-owned banks are more likely to be affected by the government (Dinc, 2005; La Porta, et al., 2002), we could expect that firms with state ownership have a larger likelihood of getting loans from state-owned banks The effect of state ownership on the probability of obtaining loans from all banks still needs empirical investigation since government guarantees are more likely through the channel of state-owned banks

The empirical results are reported in Table 1.8 The coefficient on state ownership in column (2) is 0.212 (z = 1.91), which is positive and significant, providing supporting evidence that firms with state ownership are more likely

to get loans from state-owned banks This also provides evidence for one of the channels of soft budget constraints for firms with state ownership Although the coefficient on state ownership (0.037) is positive, it is not significant, suggesting that private firms could obtain loans from non-state-owned banks Based on some unreported regression results, the bank loan interest for firms with state ownership is not significantly higher than that of other firms The reason might be that the bank loan interest is controlled by the central government during the sample time period

[Insert Table 1.8 Here]

1.5.7 Discussions on the dummy variable SOE

The dummy variable SOE is defined based on firm’s controlling shareholder Although I have shown the high correlation between SOE and state shares,

there are still some concerns Firstly, the definition of SOE using control rights

is very strict It usually requires the firm has more than 30% state shares There are some possibilities that some firms with high state ownership are classified as non-SOEs Secondly, the findings that firms controlled by the government have lower default risk might not be generalized to firms with state shares The controlling effect might be stronger than the ownership effect

In the regressions not reported in the paper, I conduct analysis directly using data on state shares before the split stock reform I find that higher state shares lead to lower default risk The state shares are absolute values, not dummy variables based on some criteria This provides evidence for the ownership effect

1.6 Conclusion

This paper is the first to directly test the association between state ownership and firm default risk I find that the presence of state ownership leads to lower default risk due to the soft budget constraints And the effect of state ownership is more significant for firms operating in competitive industries Then, I examine the effects of state ownership when the budget constraint becomes harder I find evidence that state ownership has a less significant effect for firms located in regions with less government intervention and a better legal environment These results could help us better understand the role

of government, competitions and market development To address the

potential endogeneity problem, I examine the effects of state ownership on default risk when firms are facing negative shocks to default risk I find that state ownership has a stronger effect on default risk when firms are facing global negative industry shock In addition, this paper suggests that the ownership information should be incorporated into the default forecast model,

at least in longer time-window forecasting

Bibliography 1

Acharya, V V., and N Kulkarni, 2012 “State Ownership and Systemic Risk: Evidence from the Indian Financial Sector during 2007-09.” Working paper Altman, E., 1968 “Financial Ratios, Discriminant Analysis and the Prediction

of Corporate Bankruptcy.” The Journal of Finance, 23: 589-609

Berglof, E., and G Roland, 1998 “Soft Budget Constraints and Banking in

Transition Economies.” Journal of Comparative Economics 26: 18-40

Black, F., and M Scholes, 1973 “The Pricing of Options and Corporate

Liabilities.” Journal of Political Economy, 81 (3): 638-654

Borisova, G., V Fotak, K V Holland and W L Megginson, 2012

“Government Ownership and the Cost of Debt: Evidence from Government Investments in Publicly Traded Firms.” Working paper

Borisova, G., and W L Megginson, 2011 “Does Government Ownership

Affect the Cost of Debt? Evidence from Privatization.” Review of Financial

Studies, 24: 2693-2737

Bortolotti, B., V Fotak, and W Megginson, 2010 “Quiet Leviathans: Sovereign Wealth Fund Investment, Passivity, and the Value of the Firm.” Working paper

Boubakri, N., J Cosset, O Guedhami and W Saffar, 2011 “The Political Economy of Residual State Ownership in Privatized Firms: Evidence from

Emerging Markets.” Journal of Corporate Finance 17: 244-258

Chen, G.M., M Firth, Y Xin and L P Xu, 2008 “Control Transfers, Privatization and Corporate Performance: Efficiency Gains in China’s Listed

Companies.” Journal of Financial and Quantitative Analysis 43: 161-190

Chow, C K W., F M Song, and K P Wong, 2010 “Investment and the Soft

Budget Constraint in China.” International Review of Economics and Finance,

19(2): 219-227

Christiansen, 2011 “The Size and Composition of the SOE Sector in OECD

Countries”, OECD Corporate Governance Working Papers, No.5,

www.oecd.org/daf/corporateaffairs/wp

Cull, R., and L C Xu, 2003 “Who Gets Credit? The Behavior of Bureaucrats

and State Banks in Allocating Credit to Chinese SOEs.” Journal of

Development Economics, 71 (2): 533-559

Dinc, I S., 2005 “Politicians and Banks: Political Influences on

Government-owned Banks in Emerging Markets.” Journal of Financial Economics, 77 (2):

453-479

Duan, J., J Sun, and T Wang, 2012 “Multiperiod Corporate Default

Prediction – A Forward Intensity Approach.” Journal of Econometrics, 170

(1): 191-209